Techniques are described for generating metric(s) that predict survey score(s) for a service session. Model(s) may be trained, through supervised or unsupervised machine learning, using training data such as communications from previous service sessions between service representative(s) and individual(s), and survey scores provided by the serviced individual to rate the session on one or more criteria (e.g., survey questions). The model(s) may be trained to output, based on an input session record, metric(s) that each correspond to a survey score that would have been provided by the individual had they completed the survey. The model may be a concatenated model that combines a language model output from a language classifier recurrent neural network, and an acoustic model output from an acoustic feature layer convolutional neural network. Individuals can be clustered according to the metric(s) and/or other factors, and the cluster(s) can be employed for routing incoming service requests.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A computer-implemented method performed by at least one processor, the method comprising: receiving, by the at least one processor, a session record of communications between a service representative (SR) and a customer during a service session in a service environment; providing, by at least one processor, the session record as input to at least one computer-processable model that determines, based on the session record, a prediction of at least one survey score that would likely be submitted by the customer for the service session, the at least one model having been trained, using machine learning and based at least partly on survey data for previous service sessions; in response to a determination that a completed survey was not received from the customer in response to the service session, determining, by the at least one processor, a cluster for the customer based at least partly on the prediction of the at least one survey score, the cluster including a plurality of other customers having at least one similar predicted survey score as determined by the at least one model; in response to receiving a subsequent service request from the customer within the service environment, selecting, by the at least one processor, at least one SR based on determining that an average survey score for customers in the cluster exceeds a threshold value, the average survey score being an average of survey scores that are determined, by the at least one model, based on session records of communications between the customers and the at least one SR; and responsive to identifying that the customer is included in the cluster, bypassing a default interactive voice response process by directly initiating, by the at least one processor, a subsequent service session between the customer and the at least one SR that is selected based on the cluster of the customer.
2. The method of claim 1 , wherein the cluster is further determined based on the plurality of other customers in the cluster having a similarity in one or more of: a product discussed by the plurality of other customers during service sessions; a location of the plurality of other customers; and a characteristic of the plurality of other customers.
3. The method of claim 1 , wherein the at least one SR is selected based on determining that respective customers in the cluster each have a respective survey score that exceeds a threshold value and that is determined, by the at least one model, based on a respective session record of communications between the respective customer and the at least one SR.
4. The method of claim 1 , wherein: the service session is an audio call between the SR and the customer; and the session record includes an audio record of at least a portion of the audio call.
5. The method of claim 1 , further comprising: developing, by at least one processor, the at least one model based on training data that describes, for each of a plurality of previous service sessions: a previous session record of communications during a respective previous service session; and at least one survey score provided to rate the respective previous service session.
6. The method of claim 5 , further comprising: receiving, by at least one processor, the at least one survey score provided, by the customer, to rate the service session; and employing, by at least one processor, the at least one survey score and the session record to further train the at least one model.
7. The method of claim 1 , wherein: the at least one model is trained to identify a plurality of features present in the session record; and the prediction of each of the at least one survey score is determined based on a strength of one or more corresponding features present in the session record.
8. The method of claim 7 , wherein the machine learning is deep learning in which a separate model is trained to identify each of the plurality of features.
9. The method of claim 1 , wherein the at least one model is a concatenated model that is a combination of a language model and an acoustic model; the language model is output from a language classifier recurrent neural network (RNN); and the acoustic model is output from an acoustic feature layer convolutional neural network (CNN).
10. A system, comprising: at least one processor; and a memory communicatively coupled to the at least one processor, the memory storing instructions which, when executed by the at least one processor, cause the at least one processor to perform operations comprising: receiving a session record of communications between a service representative (SR) and a customer during a service session in a service environment; providing the session record as input to at least one computer-processable model that determines, based on the session record, a prediction of at least one survey score that would likely be submitted by the customer for the service session, the at least one model having been trained, using machine learning and based at least partly on survey data for previous service sessions; in response to a determination that a completed survey was not received from the customer in response to the service session, determining a cluster for the customer based at least partly on the prediction of the at least one survey score, the cluster including a plurality of other customers having at least one similar predicted survey score as determined by the at least one model; in response to receiving a subsequent service request from the customer within the service environment, selecting at least one SR based on determining that an average survey score for customers in the cluster exceeds a threshold value, the average survey score being an average of survey scores that are determined, by the at least one model, based on session records of communications between the customers and the at least one SR; and responsive to identifying that the customer is included in the cluster, bypassing a default interactive voice response process by directly initiating a subsequent service session between the customer and the at least one SR that is selected based on the cluster of the customer.
11. The system of claim 10 , wherein the cluster is further determined based on the plurality of other customers in the cluster having a similarity in one or more of: a product discussed by the plurality of other customers during service sessions; a location of the plurality of other customers; and a characteristic of the plurality of other customers.
12. The system of claim 10 , wherein the at least one SR is selected based on determining that respective customers in the cluster each have a respective survey score that exceeds a threshold value and that is determined, by the at least one model, based on a respective session record of communications between the respective customer and the at least one SR.
13. The system of claim 10 , wherein: the service session is an audio call between the SR and the customer; and the session record includes an audio record of at least a portion of the audio call.
14. The system of claim 10 , the operations further comprising: developing the at least one model based on training data that describes, for each of a plurality of previous service sessions: a previous session record of communications during a respective previous service session; and at least one survey score provided to rate the respective previous service session.
15. The system of claim 14 , the operations further comprising: receiving the at least one survey score provided, by the customer, to rate the service session; and employing the at least one survey score and the session record to further train the at least one model.
16. The system of claim 10 , wherein: the at least one model is trained to identify a plurality of features present in the session record; and the prediction of each of the at least one survey score is determined based on a strength of one or more corresponding features present in the session record.
17. The system of claim 10 , wherein the at least one model is a concatenated model that is a combination of a language model and an acoustic model; the language model is output from a language classifier recurrent neural network (RNN); and the acoustic model is output from an acoustic feature layer convolutional neural network (CNN).
18. One or more non-transitory computer-readable media storing instructions which, when executed by at least one processor, cause the at least one processor to perform operations comprising: receiving a session record of communications between a service representative (SR) and a customer during a service session in a service environment; providing the session record as input to at least one computer-processable model that determines, based on the session record, a prediction of at least one survey score that would likely be submitted by the customer for the service session, the at least one model having been trained, using machine learning and based at least partly on survey data for previous service sessions; in response to a determination that a completed survey was not received from the customer in response to the service session, determining a cluster for the customer based at least partly on the prediction of the at least one survey score, the cluster including a plurality of other customers having at least one similar predicted survey score as determined by the at least one model; in response to receiving a subsequent service request from the customer within the service environment, selecting at least one SR based on determining that an average survey score for customers in the cluster exceeds a threshold value, the average survey score being an average of survey scores that are determined, by the at least one model, based on session records of communications between the customers and the at least one SR; and responsive to identifying that the customer is included in the cluster, bypassing a default interactive voice response process by directly initiating a subsequent service session between the customer and the at least one SR that is selected based on the cluster of the customer.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
February 27, 2018
July 14, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.